Basic PECVD Plasma Processes (SiH4 based)
PECVD SiNx:
SiHx + NHx Î SiNx (+H2)
or SiHx + N Î SiNx (+H2)
PECVD SiOx: SiHx + N2O Î SiOx (+H2 + N2)
PECVD SiONx: SiHx + N2O + NH3 Î SiONx (+H2 + N2)
PECVD a-Si:H
PECVD SiC:
© Oxford Instruments plc 2003
SiHx Î Si (+H2)
SiHx + CHx Î SiCx (+H2)
Oxford Instruments
Plasma Technology
Types of SiH4 supply
SiH4 can be supplied as either pure SiH4 or dilute in an ‘inert’ carrier gas ,
typically N2, Ar, He.
Typical percentage dilutions are 5%, or 2% or 10%.
When converting a recipe for a different SiH4 dilutions, always remember that it is
the SiH4 flow that is important:
400sccm 5%SiH4/N2 = 20sccm SiH4 + 380 sccm N2
So you will need to flow 1000sccm 2%SiH4/N2 to achieve the same SiH4 flow.
The effect of the additional N2 flow will be minimal, and can sometimes be
compensated for by reducing the separate N2 MFC flow (if applicable).
Important note: Always remember to check rotary pump purge is suitable for
maximum SiH4 flow:
Purge flow (litres/min) > 3 x max SiH4 flow (sccm) / 14
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
PECVD Trends (SiH4 based processes)
SiNx (Nitride)
↑ SiH4 flow
↑ NH3:SiH4 ratio
↑ 13MHz power
↑ pressure
↑ temperature
Dep.
rate
↑
↓
↑↑
↑↑
↓
Refr.
Index
↑↑
↓↓
↓
Dep. Rate
Uniformity
Refr. Index
Uniformity
↑↑
↓↓
↑↑
↓↓
↑↑
↓?
Film Stress
↓↓ (more compr.)
↑
↓↓
↑↑ (more tensile)
↑
BHF Etch
rate
↑↑
↑↑
↓↓↓
SiOx (Oxide)
↑ SiH4 flow
↑ N2O:SiH4 ratio
↑ 13MHz power
↑ pressure
↑ temperature
Dep.
rate
↑↑
↓?
↑
↓?
↑
Refr.
Index
↑
↓
↓
Dep. Rate
Uniformity
↓↓
↓?
↑↑
↑?
Refr. Index
Uniformity
↑↑ (more tensile)
BHF Etch
Rate
↑↑
↓↓ (more compr.)
↓↓
↓?
↑?
↑
↓↓↓
**Uniformity is measured as centre-edge
© Oxford Instruments plc 2003
Film Stress
Oxford Instruments
Plasma Technology
Refractive index – definition
Tank tracks move
slower across
different surface
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
Refractive index – why is it important in PECVD?
Refractive index is a good indicator of film composition, i.e. Si:N
ratio or Si:O ratio.
(If Si content is high, the refractive index will be high)
It can be easily measured by ellipsometer or prism coupler,
allowing rapid evaluation of film composition (and unifomrity of
composition).
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
REFRACTIVE INDEX OF SILICON DIOXIDE:
GAS FLOW (N20/SIH4 RATIO)
1.495
1.49
1.485
RI
1.48
1.475
1.47
1.465
1.46
1.455
15
25
35
45
55
N2O/SIH4 RATIO
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
65
75
85
DEPOSITION RATE OF SILICON DIOXIDE:
GAS FLOW (N2O/SIH4 RATIO) - (High rate
process)
DEPOSITION RATE (NM/MIN)
450
400
350
300
250
200
150
100
50
0
15
25
35
45
55
N2O/SIH4 RATIO
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
65
75
85
SiONx refractive index control
Refractive index vs N2O : SiH4 : ratio
1.51
With 100 s ccm NH3
1.5
Without NH3
1.48
Refractive index vs NH3 flow
1.47
1.474
1.46
1.472
1.45
1.47
30
40
50
60
70
1.468
RI
RI
1.49
N2O : SiH4
1.466
1.464
1.462
1.46
1.458
0
20
40
60
80
NH3 flow (sccm )
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
100
120
Film quality vs dep temperature
BHF rate of SiN
1600 nm/ min
Low rate
dep
1400 nm/ min
1200 nm/ min
High rate
dep
1000 nm/ min
800 nm/ min
Thermal
oxide
600 nm/ min
400 nm/ min
200 nm/ min
0 nm/ min
0 °C
200 °C 400 °C 600 °C 800 °C
Warning: buffered hydrofluoric acid (BHF) is
highly corrosive, please read safety datasheet
and safe system of work before use.
© Oxford Instruments plc 2003
BHF rate of SiO2
1000 nm/ min
900 nm/ min
800 nm/ min
700 nm/ min
600 nm/ min
500 nm/ min
400 nm/ min
300 nm/ min
200 nm/ min
100 nm/ min
0 nm/ min
0 °C
Oxford Instruments
Plasma Technology
Low rate
dep
High rate
dep
Thermal
Oxide
200 °C 400 °C 600 °C 800 °C
Film quality vs dep temperature
KOH rate of SiN
4.0 nm/ min
3.5 nm/ min
Low rate dep
3.0 nm/ min
High rate dep
KOH rate of SiO2
2.5 nm/ min
2.0 nm/ min
16.0 nm/ min
1.5 nm/ min
14.0 nm/ min
Low rate dep
1.0 nm/ min
12.0 nm/ min
High rate dep
0.5 nm/ min
10.0 nm/ min
0.0 nm/ min
0 °C
8.0 nm/ min
200 °C
400 °C
600 °C
800 °C 6.0 nm/ min
4.0 nm/ min
Warning: Potassium hydroxide (KOH) is
highly corrosive, please read safety
datasheet and safe system of work before
use.
© Oxford Instruments plc 2003
2.0 nm/ min
0.0 nm/ min
0 °C
Oxford Instruments
Plasma Technology
200 °C
400 °C
600 °C
800 °C
SiNx – FTIR traces
• ‘Standard process’ and ‘NH3 free process’
Standard SiNx
© Oxford Instruments plc 2003
NH3 free SiNx
Oxford Instruments
Plasma Technology
SiNx - NH3 free process
Advantages
Disadvantages
•Low hydrogen content
•Lower BHF etch rate
•Better at lower
temperatures (100°C)
•Lower deposition rate (7nm
vs upto 20nm/min)
•Slightly worse uniformity
•Slightly worse repeatability
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
Film Stress - Definition
Original wafer
Wafer shape after depositing tensile film
Wafer shape after depositing compressive film
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
Film Stress - Definition
The causes of film stress can be visualised by imagining what
happens when too many atoms are packed into the film - bond
lengths are pushed shorter than normal – the film tries to relax
back to its normal bond length – pushing outwards and forcing
a convex ‘compressive stress’ curvature of the wafer.
The opposite happens for tensile stress – too few atoms per
cm3 – tensile films are usually less dense than compressive
films.
The convention is: negative sign for compressive stress,
positive sign for tensile.
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
Film Stress - Definition
E
∆ ( t substrate )
σ = 2.
.
r
t film
3 (1 − υ )
2
where
: σ = film stress
∆ = change in wafer
t substrate
E = Youngs
bow,
r = radius of scan
and t film = substrate
modulus,
and film thickness
υ = Poisson
ratio
∆, r, and thicknesses must be measured in the same unit, e.g. cm or µm
Then σ will be in the same units as E (e.g. dynes/cm2 or GPa)
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
Film Stress - Definition
for silicon:
∆ ( t substrate )
σ = 2.
r
t film
2
x 6 .16 x 10
11
dynes
cm
2
(OR 61.6 GPa OR 61600 MPa)
Example: 10µm bow on 25mm radius scan, 500µm Si substrate, 0.5µm film:
( 500 )
10
σ =
.
2
0 .5
( 25000 )
σ =
4.93 x 10
© Oxford Instruments plc 2003
9
dynes
cm
2
x 6 .16 x 10
2
=
0.493GPa
Oxford Instruments
Plasma Technology
11
dynes
=
cm
493MPa
2
Mixed frequency PECVD
At 13.56MHz, ions do not respond to RF field
At 100-350kHz, ions can respond and give
ion bombardment of growing film
Mixing of High and Low frequency power allows
control over ion bombardment and hence control
over film stress and film density
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
Film stress control
0.6
0.4
Tensile
Stress (GPa)
0.2
0
-0.2 0
20
40
60
100
-0.4
-0.6
-0.8
SiNx
Compressive
-1
HF Pulse Time (%)
© Oxford Instruments plc 2003
80
Oxford Instruments
Plasma Technology
SiOxNy
Pulsed film stress control
800Plus Film Stress
0.6
F ilm S tr e s s (G P a )
SiO2
0.4
SiN
0.2
SiON
Tensile
Linear (SiN)
0
0
10
20
30
40
50
60
70
80
-0.2
-0.4
Percentage HF = 100*HF/(HF+LF)
-0.6
Where:
-0.8
HF = HF pulse time, LF = LF pulse time,
Total HF+LF pulse time typically 20secs.
-1
Compressive
0% = continuous LF, 100% = continuous HF
Percentage HF
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
90
100
Stress control SiOx
• Reduce RF power
• Adjustment of N2O:SiH4 ratio
REFRACTIVE INDEX OF SILICON DIOXIDE:
GAS FLOW (N20/SIH4 RATIO)
1.495
0 - -50MPa
1.49
1.485
RI
1.48
1.475
-300MPa (low rate)
1.47
-100MPa (high rate)
1.465
1.46
1.455
15
25
35
45
55
N2O/SIH4 RATIO
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
65
75
85
Stress control SiNx
Stress (MPa tensile)
Stress
600
500
400
300
200
100
0
0.8
0.9
1
NH3/SiH4
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
1.1
1.2
Stress control: SiNx
Stress with process pressure
150
100
50
0
450
500
550
600
650
700
Stress
Pressure (mT)
Stress (MPa)
Stress (MPa)
200
350
300
250
200
150
100
50
0
Stress
0
10
20
30
HF power (W)
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
40
50
a-Si:H
•
•
•
•
Deposited using SiH4
Either pure, He or Ar dilution
Common for addition of PH3 and B2H6 as dopant
Surface pre-cleans useful for surface adhesion
improvements
• Bubbling of film may result when depositing on to bare Si
wafers
• Usually deposited on to SiOx or SiNx underlayer
• Stress dependant on underlayer
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
a-Si: stress control: SiOx underlying film
Stress vs SiH4 flow
0
15
16
17
18
19
20
Stress (MPa)
-50
-100
Stress
-150
-200
-250
SiH4 flow
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
a-Si: stress control: SiNx underlying film
Stress vs dopant for a-Si
0
Stress (MPa)
-100
0
1
2
3
4
-200
B2H6
-300
-400
-500
-600
Dopant flow (sccm)
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
a-Si: deposition rate
Rate vs power for a-Si (300C)
Dep rate (nm/min)
20
18
16
14
Oxford 2Torr
12
10
Oxford 1Torr
8
6
4
6
8
10
12
Pow er (W)
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
14
16
Common PECVD problems
• Poor adhesion – change chemistry of, or lengthen, surface preclean step
• Surface interactions – deposit SiNx at <300C on InP to avoid rough/lumpy
films - or use NH3-free SiNx
• Particles – if seen as silica dust in showerhead pattern on wafers then
need to search for air leaks in gas lines or behind showerhead. Or increase
pump/purge cycle after chamber cleaning.
• Particles - if random scattering of particles check chamber/showerhead
condition – may need plasma cleaning or sandblasting clean.
• Pinholes – as for particles above.
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
Particle descriptions
When they most often
occur
Possible causes
Remedy/Quick Fix –Test
Small particles less than 5um which
appear in concentrated clusters.
These clusters appear in a pattern
which mirrors that of the
showerhead holes.
The first run after a clean
Running the machine too soon after the
completion of a clean process. Silane forms
particles when it reacts with residual oxygen
in the gas lines (remember all of the gas line
up to the normally open, hardware interlock
nupro valve is incorporated in the chamber
vacuum and needs to de-gas at the end of a
long clean run).
Wait 30 minutes before running a
deposition process using silane
after finishing a clean.
The first run after a long
period of machine disuse
(say overnight)
A small leak in the silane line, particularly
around the mass flow, allowing a build-up of
silane dust which is blown though on to the
first wafer.
Fix the leak in the silane line. Flow
silane gas after a significant
period of machine disuse without
a wafer in the chamber to clear
the dust.
Every run
A leak in the gas in-let assembly or a severe
leak in the silane line.
Leak check chamber and gas line.
If both less than 1mT per minute
contact Oxford service
department and give this
description. If greater than 1 mT
per minute take apart gas inlet
assembly and clean O-rings and
PTFE part.
They are concentrated mainly in
one focal plane of the microscope
and appear to be at the bottom of
the film.
Small particles less than 5um which
appear in concentrated clusters.
These clusters appear in a pattern
which mirrors that of the
showerhead holes.
They are concentrated mainly in
one focal plane of the microscope
and appear to be at the bottom of
the film.
Small particles less than 5um which
appear in concentrated clusters.
These clusters appear in a pattern
which may or may not mirror that of
the showerhead holes.
Plasma forming behind the showerhead or in
the gas inlet assembly.
They appear in many different focal
planes of the microscope, at regular
intervals throughout the film.
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
Flakes or larger non-metallic
particles
First run after a clean
Residual particles not etched during the
cleaning process
Vacuum the chamber inside, this is
necessary periodically after
cleaning. It may be a good idea to
cool the chamber first to prevent
risk of injury with the hot table.
After a power failure or
other reason which
caused a significant drop
in table temperature
When the lower electrode cools deposited film,
particularly around the edges, cracks and is
blown on to the wafer during subsequent
deposition runs.
Clean the chamber.
After a certain amount of
deposition on the
chamber, but it varies
when they occur.
If you are depositing films of many different
chemistries and stresses, particularly those
with high stress, then the film will flake off
much earlier than expected.
Clean more regularly.
After a certain amount of
deposition but it seems to
be getting less and less
after every clean.
The films are not adhering to the showerhead
very well. Someone has cleaned the
showerhead using solvent, leaving behind a
residue that is giving poor adhesion for the
deposited films. The showerhead has become
dirty and the clean process is unable to clean it
– the showerhead is ready for its periodic
maintenance.
Beadblast the showerhead.
Metal particles which shine under
normal cleanroom light and are
greater than 20um in maximum
dimension.
Mostly all the time
Showerhead holes may be lighting-up or the
showerhead holes have become damaged due
to normal wear and tear.
Beadblast the showerhead.
Particles or marks which appear
randomly on the wafer, but look as if
they are underneath the film.
Every run
The wafer has been cleaned using solvents
which have not been properly washed off with
de-ionised water.
Use a fresh wafer straight from a
new box.
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
Common PECVD problems - Poor uniformity
1.
Change process conditions as detailed in trend tables.
2.
If processing involves LF then check;
a) is the uniformity of the LF part of the deposition causing the
problem?
b) conditioning of chamber
i.
How much deposition; LF much more affected by insulation
build up in chamber?
ii.
How clean?
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
Common PECVD problems - Poor uniformity cont’d
c) contact of sample to electrode
i.
Flatness of electrode
ii.
Recessed pin or starlift (approx 1mm)? Can become raised
over time due to flakes/dust falling into starlift hole.
iii.
If batch processing – do all of the recesses on the
carrierplate have the same coating of film (i.e. the film on
the carrierplate) – previous runs may have been carried
out with a single wafer to check rate and RI. If not, then
clean carrierplate/electrode.
d) or increase LF frequency
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
PECVD interlocks
PECVD clean gases (CF4/O2) are interlocked from all
deposition gases as a safety feature to avoid reaction of O2
and SiH4. Such a reaction is a safety hazard as this is an
explosive mixture, and is bad for the process since it will form
white SiO2 dust in gas lines.
The gas lines are ‘hardware interlocked’ by 2 Nupro valves (1
normally open, 1 normally closed) in gas pod to provide
maximum safety.
However, in order to prevent dust build up in gas lines it is
recommended that at least 2 pump/purge cycles (5min
pump/5min purge 500sccm N2, 2Torr) are carried out between
cleaning and deposition or vice versa.
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
PECVD – Chamber cleaning
If the high pressure, high power clean process has been run for too long
then attack of showerhead could occur forming a brown film/powder on
showerhead. This can be removed by:
-wipe off with a dry clean room wipe
-Eventually may need to be beadblasted, but after beadblasting ensure
that the holes are clear by using compressed air and a ‘paper clip’
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
PECVD - Typical Cleaning intervals
Material
80Plus
System100
SiO2
10µm
up to 100µm
SiN
10µm
20µm
Cleaning interval is lower for 80Plus due to flaking/peeling of film
from walls as a result of frequent chamber venting.
All figures given above may need to be reduced if a tight particle
spec is required.
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
PECVD – Typical ‘wet clean’ interval
Every 500-1000µm of film it will be necessary to perform a ‘wet clean’
as follows:
1) Plasma clean chamber
2) Cool down of electrode (advisable to avoid hot surface hazard)
3) Examine chamber – vacuum any flakes/particles
4) IPA wipe of chamber walls if necessary
5) Dry wipe of showerhead, or beadblast showerhead if necessary
6) Re-install showerhead, pump/purge chamber, and condition
Beadblast using alumina powder (aluminium oxide beads) of 180 grit size or less - maybe 120. Do not use any solvents.
Clean the showerhead after beadblasting using compressed air only. Hold the showerhead up to the light to check that
none of the holes are blocked by any grit from the beadblasting. Clean out holes with paper clip or similar if blocked.
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology
PECVD – showerhead bright spots
It is quite common to see PECVD showerhead holes becoming enlarged. This is caused during high-power processing (on an
80 Plus this is typically during plasma cleaning). Any holes which have slightly sharper edges will form an intense discharge
over the hole (due to the high fields generated by the sharper edges). This can be seen as a 'bright spot' in the plasma located
over the hole during the clean process.
This can cause some erosion of the hole and widening of the hole opening (on the outlet side only). Eventually, the bright spot
burning itself out, i.e. the erosion removes the sharp edges and hence the bright spot no longer occurs at that hole. This may
happen for several holes during the initial run-up of the system, until the showerhead 'stabilises' itself.
The bright spot may also result in some black/brown polymer deposition around the holes which, can cause premature flake-off
of deposited films. It is recommended that the showerhead is bead-blasted clean to remove such residues.
The bright spots should not be observed during low power (<50W) 80 Plus deposition processes. If they are, it is recommended
that the showerhead is plasma cleaned and bead-blasted cleaned until the bright spots are eliminated. It is sometimes possible
to cure the bright spot by using a de-burring tool to clean out any machining residues from the hole in question. If bright spots
are still present then it may be necessary to obtain a replacement showerhead.
The effect of the enlarged holes on the deposition results should be minimal, since they only enlarge the outlet of the hole,
hence they do not affect the gas flow.
© Oxford Instruments plc 2003
Oxford Instruments
Plasma Technology